Modelle/Low Refusal LLM API

Aktiv2 models

Low Refusal LLM API — Sprachmodelle mit minimaler Verweigerungsrate

Sprachmodelle wie Venice und Hermes mit minimaler Verweigerungsrate für ungehinderte Recherchen, forensische Analysen und kreative Freiheit auf MuAPI.

Get API Key Try Hermes 4 405B in Playground

General low-refusal

General low-refusal models2 models

Text GenerationLive

General low-refusal

Venice Abliterated

Venice Abliterated, a general-purpose chat model with reduced refusals and a 128K-token context window.

Text in

$0.80/M in · $3.60/M out

Modell testen

Text GenerationLiveFeatured

General low-refusal

Hermes 4 405B

Hermes 4 405B from Nous Research, a large open-weights model trained for steerability and low refusal rates, with a 128K-token context window.

Text in

$0.60/M in · $2.40/M out

Modell testen

What is the Low-Refusal LLM API?

The Low-Refusal LLM API is Muapi's catalog of 2 low-refusal chat LLMs, from lightweight 8B chat models to 1M-token-context reasoning models, behind one consistent REST interface so you can compare them by changing a single model name.

Senden Sie einen Prompt, optional mit System-Prompt für Tonalität und Kontext, und rufen Sie den generierten Text ab. Typische Anwendungen umfassen offene Dialoge, kreatives Schreiben, Brainstorming und Rechercheassistenten, die detaillierten Systemanweisungen folgen. Diese universellen Chat-Modelle weisen eine minimierte Verweigerungsrate auf; sie garantieren keine völlig schrankenlose Ausgabe. Sie bleiben für die Nutzung der generierten Inhalte verantwortlich.

Hauptfunktionen

2 Models, One API

Every endpoint accepts the same request body and returns text through the shared result endpoint, so switching models is a one-line change.

Reduced Refusals

These models decline fewer requests than typical assistant models. Behavior varies by model and prompt, and is not a guarantee.

System Prompt Control

Set a system_prompt to steer persona, tone, and output format for creative, analytical, or evaluation work.

Pay Per Token

Billing follows the token counts each call returns. There is no subscription and no model hosting to manage.

Long-Form Text Generation

Use these models for chat, drafting, and step-by-step reasoning.

Acceptable-Use Policy Applies

Reduced refusals do not mean no rules. Muapi's acceptable-use policy and applicable law still apply to what you generate.

Pricing

Endpoint	Input	Output	Price	Best For
venice-abliterated	Prompt + system_prompt, max_tokens	Complete text response (poll for result)	$0.80/M in · $3.60/M out	General chat and writing with reduced refusals
hermes-4-405b	Prompt + system_prompt, max_tokens	Complete text response (poll for result)	$0.60/M in · $2.40/M out	Large-model chat and reasoning that follows your system prompt closely

Billed per token from the usage each call returns; prices are per million input and output tokens.

How to Call the Low-Refusal LLM API

Submit a JSON request with your Muapi API key, save the request ID, and poll the shared result endpoint.

1. Prompt an Venice Abliterated senden

curl -X POST https://api.muapi.ai/api/v1/venice-abliterated \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "Give me a candid pros-and-cons list for quitting my job to freelance.", "system_prompt": "You are a candid, detail-oriented assistant.", "max_tokens": 1024}'

# Response: {"request_id":"REQUEST_ID"}

Call venice-abliterated with a prompt and optional system_prompt.

2. Prompt an Hermes 4 405B senden

curl -X POST https://api.muapi.ai/api/v1/hermes-4-405b \
  -H "x-api-key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt": "Follow this system prompt exactly and critique my draft essay.", "system_prompt": "You are a candid, detail-oriented assistant.", "max_tokens": 1024}'

# Response: {"request_id":"REQUEST_ID"}

Call hermes-4-405b with a prompt and optional system_prompt.

3. Ergebnis abrufen

curl https://api.muapi.ai/api/v1/predictions/REQUEST_ID/result \
  -H "x-api-key: YOUR_API_KEY"

# Poll until status is completed, then read the generated text from outputs.

Poll the shared result endpoint until the task is completed, then read the generated text from outputs.

Häufig gestellte Fragen

What is a low-refusal LLM?

A low-refusal LLM is a chat model with reduced refusals or one trained for high steerability, so it declines fewer requests than a typical assistant. It is a description of typical behavior, not a guarantee, and results vary by model and prompt.

Which models does the Low-Refusal LLM API include?

Venice Abliterated and Hermes 4 405B. Each is a separate endpoint with its own model name in the URL.

How much does it cost?

Billing is per token, deducted after each call from the counts the model returns. Rates start at $0.60 per million input tokens; see the pricing table for every model.

Will these models answer anything I ask?

No. Reduced refusals do not mean unrestricted output, and each model's knowledge and quality still bound the answer. You are responsible for how you use the output, and requests that violate Muapi's acceptable-use policy or the law are not permitted.

Can I send images to these models?

No. Every model on this page is text in, text out.

How do I get the result?

POST to /api/v1/<model-name> returns a request_id. Poll /api/v1/predictions/{request_id}/result until the status is completed, then read the text from outputs.

See the Abliterated LLM API →20 models on the Abliterated LLM API.See the Character Chat LLM API →12 models on the Character Chat LLM API.Browse every LLM on Muapi →Compare these models against GPT, Claude, Gemini, DeepSeek and every other language model in the catalog.

Ready to build with the Low-Refusal LLM API?

Create one Muapi API key and call every model on this page with a plain REST request.

Get an API Key Try Hermes 4 405B in Playground